The present invention deforms hairs from a reference pose based on one or more of the following: magnet position and/or orientation; local reference space position (e.g., a character's head or scalp); and several profile curves and variables. In one embodiment, after an initial deformation is determined, it is refined in order to simulate collisions, control hair length, and reduce the likelihood of hairs penetrating the surface model. The deformed hairs can be rendered to create a frame. This procedure can be performed multiple times, using different inputs, to create different hair deformations. These different inputs can be generated based on interpolations of existing inputs. Frames created using these deformations can then be displayed in sequence to produce an animation. The invention can be used to animate any tubular or cylindrical structure protruding from a surface.